Deep Learning–Driven Spatiotemporal Modeling of Domestic Tourism Dynamics in Thailand

Numerous metrics, like as visitor numbers, tourism net profit, and hotel occupancy rates, are included in the dataset, which covers 77 provinces. A baseline-based concept of shock-recovery is introduced to measure the impact and the different recovery paths in different regions. Recurrent neural networks incorporate engineered elements that capture seasonality, trend dynamics, shock strength, volatility, and recovery tim-ing. Importantly, latent spatial heterogeneity and cross-regional dependencies are learned inside a single architecture by integrating province-level spatiotemporal em-beddings. To jointly forecast tourism demand and net profit, models called Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) are created. Using a time-preserving evaluation technique, model performance is assessed against statisti-cal time-series baselines and XGBoost. In early 2020, the results show a structural break that exceeded the 95% decline, along with significantly unequal recovery pat-terns. By roughly 22-28% in RMSE and 14-16% in MAPE, the suggested deep learning models surpass baselines, exhibiting superior ability to capture spatial heterogeneity and nonlinear recovery dynamics.